Introduction
The Canada-United States Safe Third Country Agreement (STCA)Footnote 1 is a bilateral treaty that significantly restricts refugee claims made from either side of the Canada-US land border. Negotiations over the eventual treaty were initiated by Canadian officials in 1992, completed in 2002, and continued in the form of a revised treaty in 2023. While initially limited to ports of entry, the revised treaty extends its provisions to the entire land border. The agreement has been extensively analyzed and criticized by scholars of refugee law because it appears to undermine Canada’s postwar humanitarian commitment to refugee protection (Arbel, Reference Arbel2013; Macklin, Reference Macklin2004; Carasco, Reference Carasco2003). Indeed, throughout the negotiation over the initial treaty across successive governments led by both liberals and conservatives, political leaders and bureaucrats faced significant pressure from interest groups and legal experts to abandon the process. For advocates and scholars alike, Canada’s tenacious pursuit of this treaty—notwithstanding these countervailing pressures—is a puzzle that stands in contradiction to its leadership on other aspects of refugee protection.
The prevailing explanation for Canada’s pursuit of the STCA is that it represents a “European turn” in Canadian refugee policy, from “policy innovator and humanitarian leader” to policy student: a “follower and adaptor of a key set of restrictionist asylum policies practiced in Europe—most notably the Dublin agreement” (Soennecken, Reference Soennecken2014: 102). Twelve European countries originally agreed to the Dublin Convention in 1990, which prevented asylum seekers from making a refugee claim in more than one state. According to these accounts the STCA is presented as an attempt to emulate the European Dublin regime as part of a broader process of policy transfer (Soennecken, Reference Soennecken2014; Macklin, Reference Macklin, Lambert, McAdam and Fullerton2013; Irvine, Reference Irvine and Skogstad2011a; Costello, Reference Costello2005; Cutler, Reference Cutler2004; Abell, Reference Abell1997). In this article, we argue that while Canadian officials did engage in policy learning from European counterparts, they were not emulating European models in their pursuit of the STCA. On the contrary, the policy process was driven by Canadian bureaucrats who sought to negotiate a tension between their pursuit of migration system integrity and upholding international law—while avoiding the negative lessons of the Dublin regime in Europe. Our findings are supported by interviews with former Canadian officials, archival materials and a systematic review of press coverage in major Canadian newspapers. This article makes a broader contribution to a growing political science literature on the role of bureaucrats as policy entrepreneurs in immigration policy making.
The STCA, as much as Dublin, has been a source of policy learning for other states who have incorporated the safe third country concept into laws and treaties (Freier et al., Reference Freier, Karageorgiou, Ogg, Costello, Foster and McAdam2021). The safe third country concept was initially developed by European states in the late 1970s and early 1980s and found justification in the soft law decisions of the UNHCR executive committee (Moreno-Lax, Reference Moreno-Lax, Guy and Weckel2015: 664; McAdam, Reference McAdam, Lambert, McAdam and Fullerton2013: 29). It denies protection to asylum seekers on the grounds that they can seek protection in another country (Freier et al., Reference Freier, Karageorgiou, Ogg, Costello, Foster and McAdam2021). To the extent that Canadian officials were learning from their European counterparts in the 1980s and 1990s, they were concerned about the impact of Dublin on increasing demand for access to Canada’s refugee status determination system. More significantly, many of the “lessons” drawn from Europe appear to have been negative, with the STCA’s development shaped by intentional divergences from Dublin. This is apparent in policy makers’ explicit decision to limit the application of the initial agreement to official ports of entry to avoid legal problems associated with the Dublin regime. Although more recently characterized as a “loophole” in the STCA, this was a “feature, not a bug” at the time the agreement was created to address a perceived limitation of the European approach, with the latter viewed as prone to ambiguity and subsequent dispute regarding which state was responsible for a claim. Relatedly, there were also clear concerns with documented cases of “chain refoulement” in Europe (where refugee claimants were referred or relocated to a third country, and then returned to face potential persecution in their country of origin). Consequently, there was a sustained effort by bureaucrats to engage with the UNHCR during the development of the agreement to ensure that it was not viewed as inconsistent with Canada’s commitments under the Refugee Convention and international obligations more broadly—namely, to guarantee access to refugee status determination and to prevent refoulement. These considerations shaped the decision making of immigration bureaucrats and the wider process that eventually led to the creation of the STCA.
The introduction of the safe third country concept accompanied the design of Canada’s Immigration and Refugee Board in the late 1980s as a way to reduce the flow of asylum claims into a robust system of administrative justice. A core concern of immigration bureaucrats was upholding international law and maintaining public confidence in the integrity and effectiveness of Canada’s refugee status determination process. While the safe third country concept was initially proposed and legislated in 1988 as a unilateral action potentially targeting many countries of transit, it was not translated into regulations. Refugee advocacy groups were unified in their opposition to the provision, and they were able to sustain their objections with political leaders. Bureaucrats subsequently narrowed the scope of their ambitions to the United States and sought a bilateral treaty to clarify the terms of an arrangement that would satisfy the UNHCR and associated international legal standards. These efforts were initiated twice in the 1990s and failed both times largely because of US policy changes. The idea of an STCA was revived again in the late 1990s, and successfully pursued by Canadian negotiators with the United States in the context of broader border cooperation following 9/11.
Policy Entrepreneurship and “Cross-pressures”
The central puzzle of this paper is an inversion of the classic problem highlighted by Hollifield (Reference Hollifield2004), Boswell (Reference Boswell2007), Freeman (Reference Freeman1995) and others—that is: why do liberal democratic governments expand immigration when electorates favour closure? In our article, we are asking why the Canadian government pursued the closure of its refugee policy in the face of interest group pressure that sought to preserve relative openness. The 2001–2002 negotiation and 2023 “modernization” of the STCA is evidence of the relative political insulation of decision making by officials, and it resists the application of Freeman’s theory of immigration policy as driven by interest groups lobbying officials to receive a more expansive policy. It also contrasts with the institutionalist argument that courts and constitutional protections restrain the restrictionist impulses of immigration decision makers (Joppke, Reference Joppke1998; Hollifield, Reference Hollifield2004). The unanimous finding of the Supreme Court of Canada in favour of the government position has decisively undermined the institutionalist argument. Even further, prominent arguments made by political scientists in the 1990s that states were self-binding through their adoption and incorporation of international human rights norms are rebutted by the development of the STCA (Bauböck, Reference Bauböck1994; Soysal,Reference Soysal1994; Jacobson, Reference Jacobson1995). Not only was the constraint of embedded liberalism disregarded in the negotiation of the STCA, but international law itself became an important tool for the state to reassert control over refugee policy.
We argue that the development of the STCA was fundamentally driven by the policy entrepreneurship of Canadian immigration bureaucrats. As other political scientists have argued, Canadian immigration bureaucrats have been uniquely insulated from political pressures, which has enabled them to drive policy innovation (Elrick, Reference Elrick2022; Ellermann, Reference Ellermann2021; Triadafilopoulos, Reference Triadafilopoulos2012). The policy entrepreneurship of bureaucrats in immigration policy making has been widely studied in the Canadian context. With the introduction of the Immigration Act of 1952, bureaucrats were able to employ broad discretionary powers over admissions to experiment with selecting immigrants based on assessments of human capital and social ties (Elrick, Reference Elrick2022). Bureaucrats have also played an important part in subsequent developments in the “federalization” of immigration policy, shaping the way shared jurisdiction has impacted the role of provinces in admissions (Paquet, Reference Paquet2015; Paquet, Reference Paquet2019). In the refugee policy domain, bureaucrats played a central role in Canada’s initial decision to not accede to the 1951 UN Convention on Refugees due to concerns this would potentially constrain flexibility in determining refugee status and managing responses to refugee flows (Molloy and Madokoro, Reference Molloy and Madokoro2017). More recently, bureaucrats have played a key role in transforming Canadian asylum policy by creating channels for the diffusion of policy approaches from other national contexts (Irvine, Reference Irvine and Skogstad2011a). Likewise, we argue that immigration bureaucrats were the primary agents driving the development of the STCA.
Ellermann (Reference Ellermann2021) advances findings about bureaucratic policy entrepreneurship by theorizing a connection between the arena of policy making and the dynamics of decision making. She contends that immigration policy making unfolds in distinctive “policy arenas” which are each characterized by a special level of insulation from actors who might shape policy positions. When immigration policy is developed within the executive arena, led by bureaucrats, decision makers experience a high degree of insulation from both popular pressure and interest group pressure—while being more responsive to international pressure. Ellermann contends that the direction of policy change (liberalization vs. restriction) is shaped by the level of insulation in a particular policy arena, and policy making in the executive arena tends to favour liberalization. In our case, however, we see a paradoxical result: policy making primarily unfolded in the executive arena, but the associated popular and interest group insulation enabled bureaucrats to pursue more restrictive policies, not more liberal ones.
Our argument is that bureaucratic decision making was shaped by “cross-pressures” associated with maintaining migration system integrity on one hand and upholding international law on the other. The former pressure promoted restricted access to Canada’s RSD regime, while the latter promoted access to it. The phenomenon of cross-pressures pushing migration policy making in contradictory directions has been analyzed by other political scientists. Hampshire describes the modern liberal state in relation to four facets that exert contradictory influences on policy making: courts and capitalism incline states towards openness, while democracy and nationalism push them towards closure (Hampshire, Reference Hampshire2013). These contradictory influences are managed by decision makers by implementing policies that appear to serve contrary ends, or “conflicting dynamics of openness and closure” (Hampshire, Reference Hampshire2013: 14). Hollifield (Reference Hollifield2004) analyzes a similar dynamic in his classic theorization of the “emerging migration state,” which is trapped in a “liberal paradox”: open to migration by economic necessity, and yet restrictive in order to manage security and political risks. The result is a liberal state that both invites and restricts migration.
Christina Boswell extends Hollifield’s insights into the migration state, while also locating the influence of cross-pressures on decision making more clearly within the executive branch of state (Reference Boswell2007). She observes that states are faced with a fundamental dilemma when it comes to their migration policies because the preconditions of their legitimacy push them to try to meet competing requirements in their immigration policies. She identifies these preconditions as security, accumulation, fairness and institutional legitimacy. Decision makers are constrained by these functional imperatives when designing their migration policy. Boswell contends that when migration policy is depoliticized—that is, when decision making is made in a political arena with high degrees of insulation, as Ellermann (Reference Ellermann2021) puts it—policy makers can more readily fulfill these four imperatives within a managed and selective immigration policy. When considering Canada’s refugee policy more specifically, we find that fairness and institutional legitimacy have been primary imperatives shaping decision making. While security concerns do feature to some degree, the main policy dilemmas experienced by decision makers were between meeting a standard of fairness according to international law while also maintaining the legitimacy and integrity of Canada’s asylum system.
Although our principal interest in this article is analytical, we also note that the cross-pressures shaping the policy entrepreneurship of immigration bureaucrats presents a case of a normative “migration policy dilemma” (Bauböck et al., Reference Bauböck, Mourão Permoser and Ruhs2022). Bauböck and colleagues propose a normative theory of policy dilemmas as a characteristic of the ethics of immigration. They call these “hard” ethical dilemmas because they involve trying to reconcile conflicting moral goals, which cannot be addressed through a particular political solution. In the 1980s, Canada developed the Immigration and Refugee Board to manage in-country asylum claims, which has been referred to as a “world class” approach to refugee status determination (RSD), and as a “Rolls Royce” or “Cadillac” system of RSD (Hamlin, Reference Hamlin2014; Dauvergne, Reference Dauvergne2003: 733; McLaren, Reference McLaren1991). The creation of this robust system of administrative justice, however, also generated an increasingly urgent imperative to control access to it. If it were to be overwhelmed with cases—so the argument goes—it would cease to function with institutional legitimacy (or, perhaps, popular legitimacy, should the expense of funding it exceed public support). For the officials confronting this dilemma the trade-off was real; continued domestic support for asylum, and refugee assistance more broadly, could be undermined by perceptions of a loss of control over borders.Footnote 2 Senior immigration bureaucrats pursued the STCA as a means of managing this particular policy dilemma: maintaining the integrity of the IRB while limiting access to it in a manner that would be consistent with international law.
In the next section of the article, we discuss data sources and methodology that informs our analysis. Following this, we provide a broad reconstruction of the historical context leading to and through the development of the STCA, tracing the genesis of the agreement back to the early development of Canada’s asylum regime. In the remaining sections, we unpack the way immigration bureaucrats balanced and otherwise navigated the competing imperatives they confronted in the context of the development of the STCA, indicating the degree to which these considerations informed, shaped and constrained the development of the first bilateral asylum coordination agreement. We conclude by reflecting on the dynamics that informed this multiyear process, what this may tell us about a key moment in the history of the “migration state” in which policy makers attempted to navigate the competing imperatives of control and humanitarian constraint, and how we might position this moment relative to contemporary efforts across the global north that have prioritized the former over the latter.
Sources and Methodology
Our analysis draws on contemporaneous print media coverage, governmental and organizational documents, commissioned studies, parliamentary debate records and reports, archival data, as well as semi-structured elite interviews. Most materials we draw on are publicly available through libraries and government websites. We identified Canadian print media coverage through keyword database searches for “safe third” from 1987 to 2002. Additional archival materials were accessed with the assistance of Library and Archives Canada and UNHCR while more recent materials were acquired through Access to Information and Privacy requests.Footnote 3 The semi-structured interviews were conducted in 2023 and 2024 with policy makers and public officials involved in deliberations at key junctures in the development of the STCA. Participants were identified and recruited through organizational contacts and subsequent snowball sampling.Footnote 4 The Canadian Immigration Historical Society assisted us in identifying and contacting retired public servants. The elite interviews informing this project span nine interviews with midlevel and senior bureaucrats, all of whom were engaged in varying aspects of the creation of the STCA; however, in some cases their involvement shifted over the course of the period under analysis. Interviews were largely conducted strategically, focusing on specific bureaucratic actors with particular insights and perspectives regarding the antecedent development and/or over the course of the political narrative.
Additionally, we engage with a range of publicly available Parliamentary and Congressional materials, including the records of the Parliament of Canada’s House of Commons Standing Committee on Citizenship and Immigration and the United States’ House of Representatives subcommittee on Immigration, Border Security and Claims. The diverse methods of data collection we draw on in tandem across this project have enabled us to advance a comprehensive and nuanced account of the development and evolution of the STCA across varying bureaucratic, political and historical dimensions. By analyzing these different sources of data in conjunction with one another, we shed light on the complex and often competing imperatives that informed the choices of policy makers located at various levels of decision making.
From Bill C-55 to the STCA
The origins of the Safe Third Country Agreement can be traced to efforts by immigration officials to introduce the safe third country concept into Canadian law during the 1980s. Conceived as a legitimate way to control the rising number of refugee claimants arriving via the United States and elsewhere, the initial legislation—passed in 1988—was never translated into regulations. It became a dead letter, until it was revived (and died) twice in 1990s as Canada sought a bilateral border agreement with the United States that would include the mutual application of a safe third country designation. The Canadian pursuit of an agreement would wait until it was included in a new post-9/11 border agreement with the United States.
The introduction of the safe third country concept into Canadian policy debates was a response to the rapid growth of in-land refugee claims in the 1970s and 1980s, after Canada formally recognized refugees as a distinct class of immigrants (Molloy and Madokoro, Reference Molloy and Madokoro2017). This in turn was informed by the relatively limited attention given to asylum as a component of the major reforms that were undertaken to Canada’s immigration system through the 1960s and 1970s (Girard, Reference Girard2013). As the number of in-land claims rose, a principal concern of officials was to address a growing backlog of cases that could undermine public confidence in the regime and its delivery of administrative justice to refugees (Girard, Reference Girard2023; Anonymous official, 2024). The designation of safe third countries (including, but initially not limited to, the United States) was proposed as a crucial measure to address this aim.
After the 1985 decision of the Supreme Court of Canada (Singh v Minister of Employment and Immigration), the government was required to reform its emerging system of refugee status determination to incorporate an oral hearing for claimants. The process of drafting legislation that would create the Immigration and Refugee Board (IRB) took three years, partly due to opposition by NGOs to the inclusion of a safe country principle in the enabling legislation. Bill C-55 became “an official’s bill” because ministers wanted to stay away from it; the bill was drafted by bureaucrats and defended by them in front of the legislative committee (Girard, Reference Girard2013; Canada, 1987).
The bureaucrats charged with drafting the bill concluded that to comply with the Singh decision while also managing claims on an efficient basis, they would have to limit the number of people coming into the system (Girard, Reference Girard2019: 4). The most effective way to do this in compliance with international law was to limit arrivals from the United States, which was the source of an estimated 70 per cent of refugee claims (Segal, Reference Segal1988). The aim of the new legislation was to prevent claims from people who had passed through a country that upheld the legal principle of non-refoulement (Canada, 1987: 1725). “Our thinking,” recalls a member of the task force, “was that if this influx could be diverted away from the tribunal to more conventional forms of immigration enforcement, a great deal of pressure would be taken off the tribunal, enhancing its ability to function without the threat of rapid increases of claims” (Girard, Reference Girard2019: 5). Bill C-55 received royal assent in July 1988.
Despite the inclusion of the safe country concept in Bill C-55, however, Minister of Immigration Barbara McDougall pre-emptively placed a moratorium on it (Montgomery, Reference Montgomery1988; Girard, Reference Girard2023). Officials with the IRB almost immediately began complaining of a case backlog, which immigration officials publicly blamed on McDougall (Malarek, Reference Malarek1989; Girard, Reference Girard2023). Within several months, Gordon Fairweather, the first chairperson of the IRB, declared that the major problem facing the new IRB was the government’s failure to implement the safe third country policy (Malarek, Reference Malarek1990). This situation reflected the significant dissonance between the resources allocated to the IRB, which appeared to have been envisioned to operate in conjunction with a safe third country agreement by those involved in the legislative process that created it, and the rapidly growing number of claimants appearing before the tribunal (Dolon and Young, Reference Dolon and Young2002: 16).
When McDougall was replaced as immigration minister by Bernard Valcourt, department officials pushed again for the implementation of the safe third country provision (Bryden, Reference Bryden1991). An alternative was floated by Fairweather. He urged the negotiation of a bilateral agreement with the United States (Bryden, Reference Bryden1991). Later that year, Valcourt acknowledged that he had begun preliminary discussions with Mexico and the United States to work towards an agreement that would control the movement of asylum seekers (Oziewicz, Reference Oziewicz1991a). He claimed that a bilateral or trilateral agreement would overcome potential Charter concerns, presumably because asylum seekers outside of Canada’s borders would not be entitled to access the IRB regime.
Another reason Canada was exploring an international agreement was the signing of the European Union’s Dublin Convention in 1990, an agreement initially between 12 European states to share responsibility for hearing asylum claims (Hathaway, Reference Hathaway1992: 71–2). Some officials expressed concern that the European Union’s convention would lead to more people coming to Canada (Oziewicz, Reference Oziewicz1991b). Others saw it as an example of international coordination from which Canada could learn (Anonymous official, 2024).
By September 1992, Canada and the United States entered the final round of negotiations on a framework for the administration of asylum procedures aimed at preventing the duplication of in-land claims. Recent changes to the US asylum regime removed political interference from refugee determinations for particular groups (for example, Guatemalans, Haitians, Salvadorans), bringing the Canadian and American systems into closer alignment—despite other procedural differences (Helton, Reference Helton1993). The Canadian government also prioritized engaging with UNHCR to assess the organization’s support for the STCA, with Valcourt making the agreement a key topic of his meeting with Sadako Ogata during the high commissioner’s mission to Ottawa in May of 1993 (UNHCR, 1993b). Before the agreement could be finalized, however, national elections in both countries led to changes in government and a suspension of negotiations.
While the Liberal Party pilloried the proposed MOU in opposition, they quickly changed position once in power. After opposing policy, they began to impose policy. Sergio Marchi, previously the immigration critic, became minister of immigration and initially put the MOU proposal on hold (Canadian Press, 1993). However, as his former deputy minister noted, “he came to appreciate, as ministers do, the pressures of the whole system and the need to have a system with some integrity” (Harder, Reference Harder2023). Within two years, he announced that a deal could be signed with the Americans during an upcoming visit to Ottawa by President Clinton, to the shock of NGOs (Thompson, Reference Thompson1995a). Marchi was persuaded by immigration bureaucrats, who urged him to sign an agreement with the United States before the Americans began processing a backlog of cases that could lead rejected claimants to head north (Irvine, Reference Irvine2011b; Thompson, Reference Thompson1995b). As part of a “Shared Border Accord” reached between President Clinton and Prime Minister Chrétien in 1995, the two countries undertook to conclude a bilateral agreement on sharing responsibility for asylum seekers (Canada, 1996a).
The government released a preliminary draft agreement for public consultation in November 1995, inviting formal comments from the UNHCR, convening a dialogue with NGOs and presenting the draft for review by the House of Commons Citizenship and Immigration Committee (Canada, 1996a). Ogata, as head of UNHCR, offered direct support for an agreement, contradicting the opposition presented by the Canadian Bar Association, Amnesty International and the Canadian Council of Churches (Harder, Reference Harder2023). According to the former deputy minister, “We could not have done what we did without the alignment of UNHCR and the Government of Canada’s policy” (Harder, Reference Harder2023). The UNHCR expressed the hope that it would “set an example for similar future accords” (Canada, 1996a). It also noted the draft agreement had been revised to take into consideration principles of international refugee law and emphasized that the agreement should ensure that state responsibility for considering refugee claims is clearly specified so that refoulement and “refugee in orbit” situations are avoided (Abell, Reference Abell1997).
The preliminary draft agreement was ultimately abandoned by the government in early 1998, however. The passage of the Illegal Immigration Reform and Immigrant Responsibility Act (IIRIRA) of 1996 by the US Congress created changes to the US asylum regime that undermined the draft agreement. It introduced summary deportation procedures that would block asylum seekers from access to RSD procedures if they failed an initial screening test (Acer and Byrne, Reference Acer and Byrne2017). According to an official involved with the preliminary draft agreement, the two governments decided that in the wake of this law they would have to set it aside and “put it on ice indefinitely” (Anonymous official, 2023). The working assumption behind the agreement was that the two countries’ asylum systems met the standards of international law, and the changes to the US system potentially undermined this assumption (Anonymous official, 2023). If Canada could not be guaranteed that a refugee claimant would have access to an RSD procedure in the United States, it could not continue pursuing the agreement.
Although the preliminary agreement was left unfinalized, the rising number of refugee claims over the next several years kept the idea of a safe third country agreement alive among bureaucrats. The number of refugee claims more than doubled from 1997 to 2001 (see Figure 1). Senior officials estimated that these numbers could be reduced by 5,000 to 6,000 a year with the introduction of a safe third country agreement (Canada, 2002a). In February 2001, the Immigration and Refugee Protection Act (IRPA) went through first reading in the House of Commons, and it included s.102(1) which allowed for the introduction of regulations to designate countries as “safe” that comply with Article 33 of the Refugee Convention and Article 3 of the Convention Against Torture. IRPA also noted that Canada should consider creating an agreement with another country “for the purpose of sharing responsibility with respect to claims for refugee protection,” prior to designating it as “safe” under the legislation (IRPA s.102(2)(d)).
Figure 1. Annual asylum claims and backlog in Canada (1989–2023)Footnote 5 .
In the wake of the attacks of September 11, 2001, US security concerns took centre stage in Canada-US relations. One of the primary American interests was to address border security, and this included making new demands on Canadian counterparts. As Peter Harder recalls, there was a higher level of political engagement with the refugee issue than was previously possible: “the higher political and policy conversation meshed into the consequence of 9-11” (Harder, Reference Harder2023). In December 2001, Tom Ridge and John Manley signed a Smart Border Declaration, which was expanded a year later into a 30-Point Action Plan that included the STCA. When called upon to explain the STCA to a Congressional Subcommittee, Kelly Ryan (Deputy Assistant Secretary, Department of State), said:
Negotiation of this proposal was undertaken after September 11th, and at the request of the Government of Canada as part of the 30-point action plan….We view this as an important agreement in the context of the overall 30 points that Canada wants and that we are willing to agree to as a trade-off for the other important counter terrorism measures…. The United States or the Executive Branch has no intention now of entering into any arrangement with another country other than Canada. We are doing this agreement, if we go through with it, at the request of Canada, because they believe it is important for reducing their asylum backlog. We don’t view this as a counterterrorism measure. (US House of Representatives, 2002)
This account is confirmed by a senior Canadian official with knowledge of the negotiations:
It was a trade-off. They wanted many things from the Government of Canada. They wanted us to impose visas on countries. They wanted us to share information with them. We said, OK, then this is what we want. We want a responsibility sharing agreement. (Anonymous official, 2023)
And Canada got it.
Bureaucratic Decision Making
We argue that immigration bureaucrats responsible for Canadian asylum policy in the 1980s and 1990s responded to two competing imperatives: promoting migration system integrity and adhering to international law. Drawing upon interviews with bureaucrats and their contemporaneous testimony before parliamentary committee hearings, alongside additional documents and reports produced by both governments and UNHCR, we discuss how these imperatives informed the views and actions of immigration bureaucrats.
Migration system integrity
The primary imperative informing the actions of Canadian immigration bureaucrats throughout the policy process leading to the STCA was the promotion of the integrity of Canada’s migration system. Their principal concern was preventing the accumulation of a backlog of refugee claimants that exceeded the capacity of the system. A secondary, but related, issue was the challenge of carrying out removals (deportations) for failed asylum seekers.
As a country bounded by three oceans and a single land border with the United States, the management of immigration to Canada has been historically accomplished through visa policy and cooperation with American officials. Prior to Canada’s 1969 ratification of the UN Convention on Refugees and its Optional Protocol, border management included a long-standing reciprocal agreement between the two countries on turn-backs. This effectively meant that removals could be carried out within about 24 hours.Footnote 6 An individual arriving through the United States who sought entry to Canada without a visa and was inadmissible would be turned back to the United States (Girard, Reference Girard2023). This operational agreement between the two countries helped to limit the use of procedures for deporting inadmissible immigrants after their arrival. However, after the United States and Canada ratified the convention and later incorporated its definition of a refugee into legislation (in 1980 and 1976), this reciprocal agreement could no longer be applied to certain migrants seeking entry at the shared border.
Throughout the 1970s and 1980s, Canada received an increasing number of asylum claimants. From several hundred a year, by 1985 there were more than 8,000, and by 1988, about 45,000 (Gibney and Hansen, Reference Gibney and Hansen2003). Canadian policy makers, however, initially viewed asylum seekers through the framework of normal immigration rather than refugee policy. The Canadian policy on asylum seekers could be reduced to non-refoulement. However, a range of deterrent measures were quickly adopted to limit demands on Canada’s immigration system (Girard, Reference Girard2023). These included visa requirements for countries that were significant sources of asylum seekers, stationing immigration officers to check boarding passengers, and later, financial penalties for airlines carrying refugee claimants and interdiction measures to prevent the arrival of human smugglers into Canadian waters (Harder, Reference Harder2023).
The introduction of the safe third country concept into Bill C-55, therefore, had several interrelated objectives. The first was to control processing backlogs from overwhelming the migration system. Processing backlogs created organizational challenges for government officials, they undermined the rights of migrants to family reunification and ultimately complicated enforcement actions for failed refugee claimants (Anonymous official, 2024). In short, they threatened the legitimacy of the migration system and potentially compromised public support for immigration programs more generally.
The second objective was to maintain public support for immigration in Canada. Indeed, this is stated in the federal government’s court filings: the STCA “enables both countries to strengthen the integrity of the institution of asylum and the public support on which it rests” (Attorney General of Canada, 2022: 1). Peter Harder, former Deputy Minister for Immigration, expanded on this further:
[Safe third country] is an important aspect of our managing the border and of asylum-seeking, and [it] has allowed us to continue to have broad public support… for high levels of immigration, high levels of refugee resettlement from the sponsorship program (the overseas portion), and an ongoing acceptance of the convention, asylum-seekers, spontaneous asylum-seekers, through the Immigration and Refugee Board. (Harder, Reference Harder2023)
Harder went on to acknowledge the challenge of balancing individual cases with “the integrity of the system,” but asserted that “sending people back is a better message than saying it’s a long process in Canada” (Harder, Reference Harder2023).
The third objective was to reduce the number of failed or fraudulent claims in the system. When the safe third country concept was first introduced, officials claimed that it would address “asylum-shopping.” While officials acknowledged that the concept was controversial, their aim was to avoid incentivizing “opportunistic movement” that would crowd out the ability of Canada’s refugee determination system to serve those people “who genuinely felt they had a protection need” (Anonymous official, 2024). As a senior official explained to the legislative committee on Bill C-55, the safe third country rule aims to limit the number of times and how many systems a person can access to advance a refugee claim. If someone passes through one country and fails to secure refugee status, and then moves on to another, “should he be able to try the next one and the one after that and the one after that?… That [first] decision should apply for all of the other countries associated” (Canada, 1987).
The pursuit of a safe third country agreement by officials was animated by a shared interest in preserving the integrity of the migration system, for which they were responsible. As Girard remarks, “Canada cannot assist everybody who needs help. The problem is bigger than any one nation can handle. We have to plan and meter our activities to serve the greatest needs. This calls for the making of choices that are always controversial and choices that exclude as well as include” (Girard, Reference Girard2023). The interest of bureaucrats in uniformly excluding arrivals via the US land border was, however, moderated by other considerations. These ultimately led to delays and exceptions to the original concept of the safe third country.
International law
If maintaining migration system integrity was the primary imperative driving interest in asylum coordination with the United States, upholding international law was of arguably equal importance to bureaucrats. In this sense, officials confronted a “two-level game” that required implementing policy solutions that would respond to perceived domestic pressure to address weaknesses in the country’s asylum regime, while also not compromising either the country’s explicit legal obligations under the refugee convention or transgressing soft law international standards.Footnote 7
The abiding concern with maintaining compliance with international law would shape both the general architecture and specific form of the STCA as a bilateral framework. Within this context it is worth noting that the 1951 Refugee Convention’s preamble observes “that the grant of asylum may place unduly heavy burdens on certain countries” while acknowledging the role of “international cooperation” as essential in the search for a “solution of a problem of which the United Nations has recognized the international scope and nature.” The convention itself does not contemplate the development of subsequent bilateral or multilateral approaches to asylum coordination or otherwise prohibit the creation of “safe third country” frameworks (Lauterpacht and Bethlehem, Reference Lauterpacht, Bethlehem, Feller, Türk and Nicholson2003: 122). In this regard, it is silent on the matter to the same degree that it articulates no formal obligation on the part of states to support resettlement in a third country. However, it does impose a series of requirements on states parties, foremost among which is the non-derogable obligation for signatories to ensure that they do not return refugees to a state where their life or freedom would be threatened. In this context the interpretive guidance of the UNHCR on these matters constitutes a crucial component of its overall supervisory role in relation to the convention and the substantive content of modern refugee law (Türk, Reference Türk2001).
Given the relatively limited modifications entailed by the 1967 protocol and reality that the dispute resolution provisions of article 38 have never been employed, the substantive requirements of international protection have largely been informed by the Executive Committee (ExCom) of UNHCR’s Conclusions on International Protection. While statements of the UNHCR’s ExCom do not in themselves constitute international law, in their capacity to “advise the High Commissioner” in the exercise of their functions under the UNHCR’s statute, these have come to constitute the primary basis for the delineation of international refugee law, informing both UNHCR practice and domestic interpretation of the refugee convention (ECOSOC, 1958; Türk, Reference Türk, Nicholson and Twomey1999). As Corkery (2006) argues, the role of ExCom’s conclusions should be understood as articulating “‘soft law’ standards and principles’” that inform the application of the 1951 convention’s provisions. The observations of Canadian immigration bureaucrats confirm this, insofar as the outputs of the UNHCR’s ExCom directly shaped the considerations of officials involved in the development of Canadian refugee policy and their approach to implementing Canada’s international protection obligations (Molloy, Reference Molloy2013).Footnote 8 In this context ExCom conclusions that supported or sanctioned the application of the safe country principle proved important for Canadian policy makers. This would initially emerge in 1993 within the context of the ExCom’s General Conclusion on International Protection, which appeared to build on the earlier statements of conclusion no. 15 (UNHCR, 1979), in noting the:
usefulness of measures to promote the prompt determination of refugee status in fair procedures, and recognizes the advisability of concluding agreements among States directly concerned, in consultation with UNHCR, to provide for the protection of refugees through the adoption of common criteria and related arrangements to determine which State shall be responsible for considering an application for asylum and refugee status and for granting the protection required, and thus avoiding orbit situations… such procedures, measures and agreements must include safeguards adequate to ensure in practice that persons in need of international protection are identified and that refugees are not subject to refoulement.Footnote 9 (UNHCR, 1993)
This was by no means an unqualified endorsement, as made clear by the expression of “major concern” with the “widespread misuse of the notion of ‘safe third country’” by the Committee to the General Assembly in 1999 (UNHCR, 1999). Although not explicitly mentioned, the likely referent of such concern was the EU’s recently implemented Dublin Regulations, which had resulted in several reported instances of “chain refoulement” and “orbit” situations. “Misuse,” however, implied qualified approval of the safe third country concept, which in principle could be understood as compatible with the 1951 convention if properly implemented.
Throughout the development of the STCA, policy makers displayed a clear sensitivity to soft law international standards and the guidance of UNHCR. For example, both the introduction of safe third country language into Bill C-55 and the later publication of draft MOUs with the United States are accompanied by formal commentary from UNHCR officials. Policy makers appeared willing to pause the development of the STCA due to concerns regarding potential changes to the US asylum system that would present questions about the standard of protection offered by the two countries (Anonymous official, 2023; USCRI, 1999). The drafters of the STCA also demonstrated responsiveness to concerns raised by UNHCR regarding the implementation of Dublin, to avoid refugee “orbit” situations and to guarantee non-refoulement. The references to the UNHCR ExCom’s conclusions in the preamble of the STCA should be understood in the context of this broader process of engagement:
Convinced, in keeping with advice from the United Nations High Commissioner for Refugees (UNHCR) and its Executive Committee, that agreements among states may enhance the international protection of refugees by promoting the orderly handling of asylum applications and the principle of burden sharing…
Although concerned with addressing what were understood to be significant challenges to the integrity of Canada’s migration regime, immigration bureaucrats were unwilling to risk compromising the country’s international obligations—despite reported proposals from lawyers for the Department of Foreign Affairs (Girard, Reference Girard2013).Footnote 10
Canada’s international legal obligations played a prominent and consistent role in the process leading up to the implementation of the STCA. As the first chairman of the IRB recalled, “I think we were trying to walk that fine line of not undermining the Convention, but in a sense, managing the flow” (Harder, Reference Harder2023). The eventual framework for the STCA reflected this balance, with Canada’s immigration bureaucrats believing that the combination of safeguards to prevent chain refoulement, provisions in Canadian immigration law requiring the periodic review of factors relevant to the safe third country designation, and a clear monitoring role for the UNHCR ensured that the agreement was fair and in line with Canada’s obligations.
Preliminary reports from UNHCR’s Branch Office in Ottawa of Canadian efforts to advance an MOU in the early 1990s reveal an initial judgment of the draft to be “generally acceptable,” which informed the broader deliberations and responses of UNHCR (UNHCR, 1993c). As the STCA moved forward, the accompanying Regulatory Impact Assessment Statement acknowledged ongoing opposition among Canadian NGOs while also stating that UNHCR “supports the objectives” of the agreement and “considers that both countries meet their international obligations” (Canada, 2004: 1625; but see also Canada, 2002b). The UNHCR’s assessments were important to policy makers, not only because of the reissuances these offered for compliance with Canada’s international obligations, but also because of the role such approval could play in legal challenges bureaucrats anticipated would be mounted against the agreement where the organization’s views would carry significant weight (Morton, Reference Morton2024), a reality also contemplated by UNHCR officials (UNHCR, 1993c). Although policy makers were aware that there were differences between the Canadian and American asylum system (and were frequently reminded of this by NGOs) they viewed both as meeting the threshold of international standards and as offering comparatively similar overall acceptance rates for refugees. The subsequent monitoring report issued by the UNHCR after the implementation of the STCA appeared to largely vindicate this view, with the agency stating that:
[I]t is the UNHCR’s overall assessment that the Agreement has generally been implemented by the Parties according to its terms and, with regard to those terms, international refugee law. Individuals who request protection are generally given an adequate opportunity to lodge refugee claims at the ports of entry (POEs) and eligibility determination decisions under the Agreement have generally been made correctly. (UNHCR, 2005)
Notably, although the monitoring report acknowledged some areas of concern, particularly regarding the “direct back policy” at ports of entry (a policy subsequently ended in response to these comments), the broader assessment of the UNHCR was clearly supportive rather than critical.
Contextual factors
The pursuit of the STCA by Canadian bureaucrats was also shaped by two important contextual factors: the Canada-US relationship and interest group pressure. The former factor prevented Canada from adopting a unilateral approach to applying the safe third concept. The latter factor delayed the finalization of an agreement until 2002 and served to reinforce the important role of UNHCR support for the STCA as a counterweight to interest group criticism.
First, Canadian immigration bureaucrats sought a bilateral agreement with the United States out of a strong desire to work cooperatively with their southern neighbour, instead of imposing new border controls unilaterally. According to an observer involved in negotiations, Canadian officials prioritized the creation of the Safe Third Country Agreement as part of a larger border accord, both in 1995 and in 2002 (Anonymous official, 2024). Indeed, the announcement of a memorandum of cooperation in late 2001 made clear that the renewal of interest in negotiating an agreement was connected to furthering collaboration on a range of issues, with Minister of Citizenship and Immigration Caplan emphasizing the two country’s “long-standing commitment to make the US-Canada border a model of cooperation” (United States, 2001). The eventual emergence of the STCA should be understood as embedded within and as a part of the US-Canada Smart Border Declaration and associated 30-Point Action Plan that preceded the agreement, which provides evidence of the broader scope of issue-linkage surrounding the negotiations (United States, 2002). The latter also provides some insight for appreciating the basis of American interest in acceding to an agreement that would seemingly provide negligible material benefits to the United States.
If developing a cooperative approach towards this cross-border issue was an important priority for both bureaucrats and their political leaders, coming to a formal agreement was also understood as a pressing necessity. Bureaucrats viewed the Canada-US land border as a primary site for refugee crossings, especially following the introduction of airline carrier sanctions. This reflected the fact that approximately one-third of asylum claims in Canada between 1995 and 2001 involved refugee claimants who were identified as having arrived from or passed through the United States (Canada, 2001b). In addressing the significant challenges this raised for Canada’s asylum system, negotiating an agreement with the United States was also related to the desire to adhere to soft international law, insofar as securing formal American guarantees of access to RSD was essential to avoiding “refugee-in-orbit” problems (Anonymous official, 2023).
Second, bureaucrats ran into pronounced and persistent countervailing pressure by advocacy groups, or NGOs. During the final decades of the 20th century, NGOs exerted a significant amount of control over the government’s refugee policy agenda (Van Kessel, Reference Kessel2013). These groups included the Canadian Council of Churches (CCC), the Canadian Council for Refugees, Amnesty International and the Canadian Bar Association. While the safe third country principle enjoyed some support from unorganized public opinion, advocacy groups sustained a campaign of opposition through media engagement, lobbying and litigation. Their criticism primarily focused on disagreement with bureaucrats over whether the United States was, indeed, a safe country for refugees. While they saw many shortcomings in the Canadian system of refugee status determination, they viewed it as superior to what was available in the United States.
When the safe third country concept was introduced into Bill C-55, advocacy groups lobbied against its inclusion and later implementation (Thompson, Reference Thompson2010). CCC representatives offered testimony that was generally agnostic on the concept itself but opposed its application to the United States (Canada, 1987: 1115). Among the main sources of concern was the differential treatment of refugee claimants from Guatemala and El Salvador, where rates of recognition were significantly lower in the United States (not mentioned was the fact that Chinese and Iraqi claimants fared better in the United States) (Harper, Reference Harper1988). After C-55 passed, and prior to its implementation, the CCC was successful at gaining an audience with Minister of Immigration Barbara McDougall. She hosted a retreat with several groups and emerged from the gathering firmly opposed to the implementation of the safe third concept gained in the legislation, declaring that the government has “postponed the decision for now” on the concept (Malarek, Reference Malarek1988). According to one official, she had an underlying mistrust of the Americans and was “plugged into the [CCC] Inter-Church Committee for Human Rights in Latin America” (Girard, Reference Girard2023).
The same groups continued to oppose the concept when it was reintroduced in a draft MOU with the United States in 1992, and again when it was incorporated into a preliminary agreement proposed as part of a broader Canada-US Border Accord in 1995. The preliminary agreement was submitted to the House of Commons standing committee on citizenship and immigration, where hearings were held with refugee advocacy groups (Canada, 1996b). Bureaucrats commented that while they were in regular dialogue with the NGOs, they did not feel that the groups had a high level of trust in government officials (Anonymous official, 2024; Girard, Reference Girard2013). Whatever level of trust they felt about Canadian officials, it was lower for American officials:
They were prepared to spend time telling me about all the flaws of our system [but] the thought of denying someone access to our system by turning them back and putting them into the American system, suddenly our system became much more attractive to them. (Anonymous official, 2024)
While advocacy groups pointed to discrepancies between the acceptance rates of the two country’s systems, Canadian bureaucrats believed that they could each be regarded as “fair” systems of refugee status determination. The arguments made by advocacy groups that the US system was “unfair” became more successful after the US Congress passed the Illegal Immigration Reform and Immigrant Responsibility Act of 1996. The legislation led to a significant increase in deportations in the United States, and a reduction in due process rights for asylum seekers. All of a sudden, the US system did seem unfair in comparison to Canada’s. Minister of Citizenship and Immigration Lucienne Robillard announced in February 1998 that negotiations over the preliminary agreement had ended “given the implementation challenges arising from…changes in the US asylum system” (USCRI, 1999). However, she left the door open to a future bilateral agreement “with appropriate protection safeguards” and implied that Canada would pursue other avenues of responsibility sharing principles contained in Canadian law.
One of the primary ways in which bureaucrats responded to advocacy group pressure was to demonstrate compliance with international law. They assiduously sought comment from the UNHCR and revised draft agreements accordingly to address concerns and mitigate risks of refoulement (Dolan and Young, Reference Dolon and Young2002: 17). In addition, the final agreement included the introduction of a monitoring mechanism by both NGOs and the UNHCR to invite third party oversight of implementation. While bureaucrats attested that they could not satisfy the NGOs, they could find approval from the UNHCR—and this helped to insulate them from broader criticism.
Conclusion
Although the STCA has now been in force for over twenty years, having survived multiple legal challenges from advocacy groups that resisted its initial introduction, the agreement has also continued to prove a source of public controversy. On the one hand, concerns regarding the agreement’s impact on Canada’s obligations towards refugees received renewed attention in the wake of the significant changes to US immigration and asylum policy undertaken during the first Trump administration. On the other hand, the agreement was viewed as contributing to a sudden rise of irregular border-crossings, with this attributed to the “loophole” in the STCA limiting its application to ports of entry, thereby incentivizing asylum seekers to cross elsewhere. The latter context led to heightened concerns regarding migration system integrity and calls to “modernize” the agreement, which eventually culminated in the sudden announcement of a new protocol to the STCA that expanded its application to the entire Canada-US land border in 2023.
Understanding these more contemporary dynamics, which are directly shaped by the bilateral architecture of the STCA, requires us to appreciate the complex factors that informed the decision making of immigration bureaucrats that guided the development of the agreement. Accordingly, this article offers an alternative explanation to arguments put forward by prior scholars that the STCA represents the Europeanization of Canadian asylum policy, or a paradigm shift in refugee policy more broadly. As we demonstrate, bureaucrats pursued various approaches to implementing the safe country principle over fourteen years of policy making across several different governments. They did so in response to a perception of growing public concerns about migration system integrity in the wake of significant changes to Canada’s immigration regime, viewing the creation of the STCA as a crucial means for preventing a backlog of cases before the IRB and sustaining public confidence in the immigration system. The shape of the agreement was influenced by the responsiveness of bureaucrats to soft international law, an interest in international cooperation and the impact of interest group pressure.
This account also provides a helpful perspective—by way of contrast—for understanding the more recent dynamics that have characterized efforts by Canadian officials to negotiate the modernization of the STCA. In particular, the latter differed in significant ways from the process that had led to the formation of the agreement. In the case we have analyzed, the extended negotiations leading up to the development of the STCA were subject to public scrutiny, and the decisions of policy makers were evidently shaped and significantly responsive to varying stakeholders and countervailing pressures. We cannot offer a definitive explanation of the divergences from this earlier path that informed the renegotiation of the STCA. One might speculate that broader developments, such as the trend toward centralization of decision making in the prime minister’s office or the long-term impact of judicializing on policy making, may have influenced these dynamics, though it is hard to know with certainty given limited available documentation during this more recent period. However, reflecting on the past process that informed the creation of the STCA offers a helpful heuristic for assessing the key factors that may have informed the decision making of bureaucratic actors in this more recent context. Understanding the broader historical trajectory of the STCA should provide resources for future scholarship in the likely event that asylum policy continues to prove an increasingly contentious subject of public debate.
Acknowledgements
Our thanks to Eline de Rooij and the three anonymous reviewers for thoughtful and valuable feedback. We would like to also express our thanks to the various officials who agreed to speak with us in the context of the interviews that informed this project, as well as to the Canadian Immigration Historical Society (CIHS), Library Archives Canada (LAC) and the UNHCR Records and Archives Section (RAS) for the support they provided. Finally, we thank Phil Triadafilopoulos and other participants from the 2023 Canadian Political Science Association panel where this research was presented. This article draws on research supported by the Social Sciences and Humanities Research Council.
Competing interests
The author(s) declare none.
Open access
